-
-
Notifications
You must be signed in to change notification settings - Fork 43
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Using custom thrust repo #105
Comments
See #38 (comment). Won't fix if you want to upgrade the bundled version to newer version. In that case, you need to roll out your own |
You can refer to https://github.com/bazel-contrib/rules_cuda/blob/cf17ea98fc73314d18cd051a848e7c9df0a97325/third_party/thrust.BUILD as an example, without forking this repo. |
The |
The root cause is https://github.com/garymm/rules_cuda/blob/d66063672a7e3ea397f40a841b2411e0e4d1ef11/examples/WORKSPACE.bazel#LL33C11-L33C11 You owe me a cup of coffee for debugging this. 😈 |
Removing |
Because there is |
If I change it to You can see the behavior here: https://github.com/garymm/rules_cuda/tree/3b973fa/examples So I think @rules_cuda//:cuda_headers should not include the thrust and cub headers. Would you be open to a PR removing them from the cuda_headers target? |
Reasonable. Would you like to contribute a PR for this? All you need to change is exclude thrust and cub from the glob and add two seperate libraries for cub and thrust. rules_cuda/cuda/runtime/BUILD.local_cuda Lines 20 to 30 in 00a1c59
|
I just tried that and it doesn't work for some reason... will investigate. |
OK, this will be tricky because the cuda toolkit include will be symlinked by bazel into the sandbox. Then all sub directory will be accessible... |
I updated the issue description with the latest status. When I run the exact command printed out by |
Hmm, indeed there's something about the sandbox that changes the behavior. Building with |
And for some reason removing |
I can contribute an example that uses custom thrust and cub with no |
I'm trying to use the latest Thrust release and running into issues.
I can reproduce this in the examples of this repo. See here: https://github.com/garymm/rules_cuda/tree/0b47488/examples
It seems that for some reason the CUDA toolkit's thrust is used, even though the
-isystem
for my custom thrust comes first on the nvcc command.To reproduce:
produces:
This is on Ubuntu, with CUDA 12.1 installed here:
The text was updated successfully, but these errors were encountered: